‘Ultra-processed’ foods, the answer to the obesity epidemic or a term to be canned?

The term ‘ultra-processed’ is becoming increasingly popularised, driven by headlines warning the public to avoid ready meals, confectionery and industrially produced bread to reduce risk of cancer (1). A recently published study Juul et al. also proposed a link between the high consumption of ultra-processed foods and risk of obesity (2). From discussing their findings within the wider literature, it was concluded that there is insufficient high-quality evidence to support a causative link between ultra-processed food consumption and obesity, and that nutritional composition of ultra-processed foods is more important than the extent of processing. Furthermore, it has been suggested that the NOVA classification system should not be used to provide dietary guidance but that focus should remain on promoting the messages underlying the EatWell Guide of  increased consumption of fruits and vegetables, wholegrains, low fat dairy products and lean sources of protein and minimal intake of discretionary (high fat, sugar and salt) items (3). This advice, and development of food products that facilitate adherence, is more likely to reduce prevalence of obesity than demonising the act of processing foods.

In the UK, it is estimated that 35% of adults are overweight and 26% are obese (4). Obesity is an important risk factor for many non-communicable diseases including diabetes, cardiovascular diseases (CVD), osteoarthritis and other musculoskeletal disorders, and some cancers (5). Consequently, 617,000 hospital admissions in the UK in 2016/17 were related to obesity (6). The simple cause of obesity is positive energy balance, where energy intake exceeds expenditure from biological processes and physical activity (5). There is considerable research surrounding the genetic link with risk of obesity, yet it cannot be denied that the modern-day obesogenic food environment, with wide availability of ‘unhealthy’ foods, has a significant role.

Conventionally, foods have been classified as food groups, such as ‘cereals and cereal products’, ‘fruit and vegetables’ or ‘dairy’. However, most of these categories poorly distinguish between healthful and non-healthful foods (7). For example, there are great variations in the nutritional quality of wholegrain breakfast cereals compared to their non-wholegrain counterparts that are coated in sugar or chocolate. The growing evidence for the detrimental health effects of processed foods, such as trans fats formed in the hydrogenation of vegetable oils increasing risk of CVD (8) and processed meat being implicated in colorectal cancer (9), has resulted in a new classification being increasingly discussed (7).

The NOVA classification system groups foods based on the extent of processing. Group 1 foods are unprocessed or minimally processed, meaning they are generally whole foods such as meat, grains, legumes, nuts, and fruit and vegetables. They may undergo basic processes such as cleaning, chilling, drying and pasteurisation for the purpose of increasing safety and palatability (10) and extending shelf-life, facilitating preparation or, for yoghurt or coffee beans, to modify flavour (11). Group 2 foods are culinary food ingredients such as starches, oils or fats, salt, sugar or sweeteners, and milk or soy proteins, tending to be unpalatable eaten in isolation but used in home or restaurant cooking and industrial preparation of foods. (10). Group 3 foods are processed foods, which are products produced by simply adding group 2 substances, such as sugar or salt, to group 1 foods, primarily for preservation or cooking. Therefore, this category includes canned fruit or vegetables, salted nuts, cured meats, cheese and artisanal bread (12). Finally, group 4 foods are those that involve processing of typically ≥5 ingredients, including additives such as stabilisers or preservatives, as well as use of industrial processes that cannot be replicated in a domestic environment (12). These foods are termed ‘ultra-processed’, generally being ready-to-eat or convenience foods that have a long shelf life and are highly palatable (10). The category encompasses numerous high fat, salt and sugar items such as soft drinks, confectionery, ice cream, sweet bakery items, pizza, burgers and other ready meals, but also includes industrially manufactured bread, infant formulas and health or slimming products (12).

According to survey data, ultra-processed foods contribute 25-50% of daily energy intake in middle to high income countries including US, Canada, Brazil and various European nations (13). The term ‘ultra-processed’ is becoming increasingly highlighted in relation to disease risk, primarily obesity, diabetes and cancer (11). A recently published study by Juul et al. investigated the association between ultra-processed foods and obesity in the US population (2). This review will discuss their results within the wider literature to determine whether ultra-processed foods are a ‘world crisis’ (7) or if the NOVA classification should be disregarded as an appropriate method of food classification when providing public health advice.

Method

Data sourcing and collection

Data from the US representative National Health and Nutrition Examination Survey (NHANES) 2005-2014 was used. Self-reported demographic, socio-economic and health behaviour data was collected, and anthropometric measurements taken. Two 24-hour dietary recalls performed 3-10 days apart were also completed.

Exposure variable and classification of foods as ultra-processed

All food items recorded were classified as ultra-processed or non-ultra-processed foods based on the NOVA classification based on a developed protocol. Information on energy value of foods was obtained from the US Department of Agriculture Food and Nutrient Database for Dietary Studies (FNDDS), with data for hand-made recipes being a sum of the individual ingredients. Relative contribution of ultra-processed foods to total daily energy intake was calculated per subject for dietary data obtained at the first 24-hour recall. This was used to divide subjects into quintiles based on ultra-processed food consumption.

Health outcomes and covariates

Obesity was measured using body mass index (BMI) and abdominal obesity. BMI ≥25kg/m² was ‘overweight and obesity’ and ≥30kg/m² was ‘obesity’, and abdominal obesity was waist circumference (WC) ≥102cm for men and ≥88cm for women.

Several covariates were of interest. This included age, sex, ethnicity, marital status, education and family poverty income ratio. Physical activity was categorised as low, medium and high based on amount of moderate intensity physical activity performed per week, and smoking status classed as never, former and current smoking.

Subject selection

Data for adults aged 20-64 years was selected, with individuals with complete data for BMI, WC, covariates of interest and at least one dietary recall being included. Participants that were pregnant or lactating, underweight and with implausible energy intake were excluded.

Statistical analysis

Participant characteristics across quintiles of ultra-processed food consumption, including dietary and behavioural aspects, were assessed, as were associations between relative energy intake of ultra-processed foods and obesity measures. The reference quintile was taken as the one with lowest relative contribution of ultra-processed foods to energy intake. Analyses were performed for the full sample and stratified by sex.

Adjustment for covariates was made, firstly with adjustment for age, and then with a model adjusted for all covariates previously mentioned. These multivariable analyses were also performed with adjustment for total energy intake. Interactions between ultra-processed food consumption and sex, age and age group were tested and sensitivity analyses were performed stratifying by total energy intake and using energy intake from processed foods as a continuous variable.

NHANES weighting factors were applied to all analyses to account for sampling bias and non-response, making the data representative of the population.

Results

Results of participant characteristics

Data for 15,977 subjects was included in the analysis. Prevalence of ‘overweight and obesity’, ‘obesity’ and abdominal obesity were 69.2%, 36.1% and 53.0%. On average, consumption of ultra-processed foods accounted for 56.1% of energy intake, with the highest to lowest quintile ranging from 84.5% to 25.4%.

Those in the lowest quintile were younger and more likely to be female, and subjects in the highest quintile had highest BMI, WC, prevalence of ‘overweight and obesity’, ‘obesity’ and abdominal obesity, and higher daily energy intake. Relative contributions to energy from carbohydrates, sugar, saturated fat (SFA) and polyunsaturated fat (PUFA) were also higher, and protein and fibre lower.

Results of health outcome analyses

Significant associations between consumption of ultra-processed foods and all obesity measures were observed in age-adjusted and fully adjusted models. For the multivariable model, the highest quintile of ultra-processed food consumption was associated with a 1.61kg/m² higher BMI and 4.07cm higher WC than the lowest quintile, and ORs of 1.48, 1.53 and 1.62 for ‘overweight and obesity’, ‘obesity’ and abdominal obesity. Linear trends were observed for each outcome.

Results of interaction and sensitivity analyses

Significant associations between quintile of ultra-processed food consumption and BMI, WC, odds of ‘overweight and obesity’ and odds of abdominal obesity were observed for women in the second to fifth quintile, and with odds of ‘obesity’ for women in the third to fifth quintile. For men, associations were significant for BMI, WC, odds of ‘obesity’ and ‘abdominal obesity’ for those in the top quintile only, and associations with odds of ‘overweight and obesity’ were non-significant. There was no significant interaction when considering age or age group.

In the sensitivity analysis, adjustment for energy had no effect on the results.

Discussion

Juul et al. concluded that an increased risk of obesity was positively associated with consumption of ultra-processed foods (2). This result is consistent with a previous ecological study where comparison of household availability of ultra-processed foods in 19 European countries with prevalence of adult obesity found those countries with highest ultra-processed food availability to have higher obesity prevalence (14), and the observation that overweight and obesity prevalence in Sweden increased between 1960 and 2010, during which time ultra-processed food consumption increased by 142% (15). Similar findings were reported by Canella et al. and Louzada et al., who used food purchasing data from the National Household Budget Survey to determine ultra-processed food consumption and assessed the association with measures of obesity (16, 17). However, it is important to note that ecological studies are unable to accurately identify cause and effect. Moreover, limitations exist in the use of household food purchase surveys as food wasted is not accounted for, nor items purchased and consumed out of the home, which are often ultra-processed in the case of takeaways and fast food, and an assumption is made that all individuals in a household have the same dietary pattern. Consequently, household surveys are an inaccurate proxy of individual food consumption. Nonetheless, a study using data from the UK National Diet and Nutrition Survey (NDNS) Years 1-4 found that a greater intake of minimally processed foods and processed ingredients (culinary ingredients) combined was associated with lower odds of overweight and obesity, although there was no association with ultra-processed foods (18). Limitations also exist within these results due to the cross-sectional nature of the NDNS, with dietary data also only being reflected by a 4-day food diary. As a result, some subjects were likely to be misclassified for exposure and reverse causality may exist. For this same reason, the quality of the study by Juul et al. could be questioned as the NHANES is cross-sectional, with data from two 24-hour dietary recalls unlikely to provide a full representation of habitual diet.

To date, there is only one prospective cohort examining associations between ultra-processed foods and obesity. Mendonça et al. studied 8451 Spanish university graduates for an average follow up time of 9 years, with dietary intake assessed by a food-frequency questionnaire completed at baseline. Over the follow-up period, a 26% higher risk of developing overweight or obesity was observed for those in the highest compared to the lowest quartile of ultra-processed food consumption (19). These results are aligned with those previously discussed, yet more prospective studies are required to ascertain whether a true association exists.

Despite the study limitations, ecological and cross-sectional surveys can provide some insight into the nature of diets high in ultra-processed foods. It is often reported that the highest quartiles of ultra-processed food consumption tend to be associated with higher total energy density, higher relative contribution to energy intake of carbohydrates, fat, saturated fat (SFA) and free sugars, higher sodium, and lower protein, fibre and potassium intake (20, 21). This was similar to the study by Juul et al., although in this instance the difference in SFA intake was minimal (2). Most notably in relation to topical UK public health concerns, Rauber et al. found an increase in free sugar of 85% across extreme quintiles of ultra-processed food intake when studying data from NDNS Years 1-6 (21), and Monteiro et al. found that 90% of added sugar consumed by the NHANES 2009-2010 subjects came from ultra-processed foods such as soft drinks, cakes, sweet snacks, breakfast cereals and ice cream, with average added sugar content in ultra-processed foods being approximately 8-fold higher than processed foods and 6-fold higher than unprocessed and processed culinary ingredients combined. The 1% increase in energy from added sugar per 5% increase in energy from ultra-processed foods also observed (22) suggests a fundamental role of ultra-processed foods in meeting the aims for sugar reduction within the UK population.

The 2015 SACN report Carbohydrates and Health advised that population intake of free sugars should not exceed 5% of total energy. However, this recommendation was primarily based on comprehensive evidence indicating an association between excess free sugar intake and both dental caries and risk of type 2 diabetes. Whereas, evidence for the link with obesity was proposed to result from inadequate compensation for energy delivered as sugar, as shown in randomised controlled trials (23). This suggests that palatability and satiety may be mediators in the link between consumption of sugary foods and obesity. In relation to ultra-processed foods, additives such as flavour enhancers can make such foods highly palatable (24), and they are often sold in large portion sizes and convenient, ready-to-eat forms, facilitating snacking and overconsumption (25). Additionally, analysis of a selection of foods commonly consumed in the UK found that glycaemic impact, measured as glycaemic glucose equivalent (GGE), was positively correlated with amount of processing and inversely correlated with satiety index (26). Furthermore, although there are great interindividual differences in taste perception and preference, innate liking for sugar to some degree is universal (27), and a habitual high fat diet has been thought to reduce gastrointestinal satiety response to fatty acids (28). Finally, the high fat and sugar content of ultra-processed foods means that a diet high in ultra-processed foods has been reported as over twice as energy dense as those containing greater proportions of unprocessed or minimally processed foods (24). As gastric satiety is highly dependent on meal volume (29), this is likely to contribute further to under compensation for energy consumed. These mechanisms offer some explanation to support a causal link with obesity.

Despite this, it is important to consider the definition of ‘ultra-processed’. It has frequently been observed that industrially manufactured bread is the most consumed ultra-processed food (17, 21, 22), yet wholemeal bread consumption is promoted due to its wholegrain content, providing dietary fibre, B vitamins, antioxidants and minerals. Importantly, dietary fibre modulates appetite by causing stomach distention, delayed gastric emptying and production of short chain fatty acids in the colon from fermentation, stimulating release of anorexigenic hormones such as GLP-1 (29). Moreover, when studying GGE and satiety index of various food items, Fardet found both wholemeal bread and all-bran cereal, deemed ultra-processed, to have a lower GGE per standard serving than white rice, which is considered unprocessed. Similarly, their satiety index was higher (26). Furthermore, in the UK such products are often fortified with essential micronutrients. For example, according to the results from the latest NDNS report, 57% of women of childbearing age have serum folate below the clinical threshold for deficiency (30), yet breakfast cereals are an important source of folate, contributing on average 12% of total intake (31). However, the definition of ultra-processed foods also rebukes synthetic vitamins and minerals but, in this instance, they are fundamental in reducing risk of neural tube defects.

The current drive for reformulation is also one that is facilitated by food processing. Replacement of nutrients detrimental to health, notably fat, salt and sugar, with ingredients designed to replicate their taste, texture and functionality without additional expense (32) is becoming increasingly important within the food industry. The aim of such product development is to facilitate behavioural change within consumers, allowing them to choose foods that have a high organoleptic quality whilst adhering to public health nutrition recommendations. It could therefore be questioned whether the definition of ultra-processed foods is one that accurately indicates nutritional quality or if sub-categorisation would be essential if this classification were to be adopted.

Finally, a study using data from Year 1 of the UK NDNS observed an inverse association between relative contribution to energy intake of ultra-processed foods and cooking ability (33). According to the recent Food Standards Agency Food and You report, respondents living in more deprived areas were more likely to consume takeaway foods and fast foods compared to those in less deprived areas (34). Juul et al. also found lower consumption of ultra-processed foods for those with higher income (2) which could indirectly relate to cooking abilities, but may also reflect the cheaper price for which many of these products are sold. Consequently, ultra-processed foods may be fundamental in ensuring certain groups of people in the UK are able to access sufficient food, although it should be noted that many of the items classified under this definition are not constituents of a healthy diet, hence these individuals would still be considered food insecure to some degree. 

Impacts

From discussing the results of the study by Juul et al., that concluded higher ultra-processed food consumption was associated with increased risk of overweight, obesity and abdominal obesity (2), it has been determined that there is insufficient supportive evidence. Notably, the quality of the evidence relating to ultra-processed foods is poor, with the literature being dominated by ecological and cross-sectional studies which cannot show causation. More prospective studies are required to enhance the evidence base.

However, it has been highlighted that there is concern regarding the nutritional quality of many ultra-processed foods, with them being energy dense and contributing to high fat, SFA, sodium and free sugar intake, nutrients that are implicated in obesity and chronic diseases. Therefore, when considering the category as a whole, it would be justified to promote replacement of ultra-processed foods with minimally processed products, which tend to have a higher nutrient density and provide the recommended dietary balance of macro- and micronutrients. Nonetheless, there seems to be no evidence that indicates ultra-processed foods such as wholemeal bread and wholegrain breakfast cereals should not be recommended as healthier alternatives to their refined counterparts or as components of a healthful diet and, in fact, their nutritional composition means they confer benefits in relation to appetite control and obesity.

Furthermore, development of less healthy ultra-processed food products in a manner that increases their protein, fibre and micronutrient content, whilst decreasing fat, SFA and free sugars, could facilitate individuals to adhere to population nutritional recommendations. This would have the greatest effect on those living in deprived areas or that have poor cooking skills, hence the UK health divide could also be reduced.

To conclude, this review suggests that utilisation of the NOVA food classification would not be beneficial for public health as it does not directly indicate nutritional quality. Additionally, the differential nature of foods within the ultra-processed foods category means it would be no less confusing than current guidance and therefore, public health recommendations should continue to focus on promoting minimisation of energy dense high fat, sugar and salt foods and increased consumption of fruits and vegetables, wholegrains, low fat dairy products and lean sources of protein, rather than demonising food processing, which is designed to ensure safety, convenience and palatability of products for consumers.

1. Gallagher J. Ultra-processed foods 'linked to cancer'. 2018. [Available from: https://www.bbc.co.uk/news/health-43064290].
2. Juul F, Martinez-Steele E, Parekh N, Monteiro CA, Chang VW. Ultra-processed food consumption and excess weight among US adults. British Journal of Nutrition. 2018;120(1):90-100.
3. FSA. The EatWell guide. Helping you eat a healthy and balanced diet. 2016. [Available from: https://www.food.gov.uk/sites/default/files/finaleatwellguide23mar2016nothernireland23rd.pdf].
4. Baker C. Obesity Statistics. 2018.  Contract No.: Briefing Paper Number 3336.
5. Wilding J. Are the causes of obesity primarily environmental? Yes. BMJ : British Medical Journal. 2012;345.
6. NHS Digital. Statistics on Obesity, Physical Activity and Diet - England, 2018 [PAS] - NHS Digital 2018 [Available from: https://digital.nhs.uk/data-and-information/publications/statistical/statistics-on-obesity-physical-activity-and-diet/statistics-on-obesity-physical-activity-and-diet-england-2018].
7. Monteiro CA, Cannon G, Moubarac J-C, Levy RB, Louzada MLC, Jaime PC. The UN Decade of Nutrition, the NOVA food classification and the trouble with ultra-processing. Public Health Nutrition. 2018;21(1):5-17.
8. SACN. Update on trans fatty acids and health. London: TSO; 2007.
9. WCRF. Meat, fish & dairy. 2018. [updated 2018-04-24. [Available from: https://www.wcrf.org/dietandcancer/exposures/meat-fish-dairy].
10. Monteiro CA, Levy RB, Claro RM, Castro IRRd, Cannon G. A new classification of foods based on the extent and purpose of their processing. J Cadernos de Saúde Pública. 2010;26:2039-49.
11. Louzada MLdC, Levy RB, Martins APB, Claro RM, Steele EM, Verly Jr E, et al. Validating the usage of household food acquisition surveys to assess the consumption of ultra-processed foods: Evidence from Brazil. Food Policy. 2017;72:112-20.
12. Parra CAM, Geoffrey C, Renata L, Jean-Claude M, Patricia J, Ana Paula M, et al. NOVA. The star shines bright. World Nutr J. 2017.
13. Mathilde Touvier TF, Bernard S, Laury S, Emmanuelle K-G, Benjamin A, Caroline M, et al. Consumption of ultra-processed foods and cancer risk: results from NutriNet-Santé prospective cohort. BMJ. 2018;360(k322).
14. Monteiro CA, Moubarac J-C, Levy RB, Canella DS, Louzada MLdC, Cannon G. Household availability of ultra-processed foods and obesity in nineteen European countries. Public Health Nutrition. 2018;21(1):18-26.
15. Juul F, Hemmingsson E. Trends in consumption of ultra-processed foods and obesity in Sweden between 1960 and 2010. Public Health Nutrition. 2015;18(17):3096-107.
16. Canella DS, Levy RB, Martins AP, Claro RM, Moubarac JC, Baraldi LG, et al. Ultra-processed food products and obesity in Brazilian households (2008-2009). PLoS One. 2014;9(3):e92752.
17. Louzada MLdC, Baraldi LG, Steele EM, Martins APB, Canella DS, Moubarac J-C, et al. Consumption of ultra-processed foods and obesity in Brazilian adolescents and adults. Preventive Medicine. 2015;81:9-15.
18. Adams J, White M. Characterisation of UK diets according to degree of food processing and associations with socio-demographics and obesity: cross-sectional analysis of UK National Diet and Nutrition Survey (2008-12). International Journal of Behavioral Nutrition and Physical Activity. 2015;12.
19. Mendonça RdD, Pimenta AM, Gea A, de la Fuente-Arrillaga C, Martinez-Gonzalez MA, Lopes ACS, et al. Ultraprocessed food consumption and risk of overweight and obesity: the University of Navarra Follow-Up (SUN) cohort study. The American journal of clinical nutrition. 2016;104(5):1433-40.
20. Mendonça RdD, Lopes ACS, Pimenta AM, Gea A, Martinez-Gonzalez MA, Bes-Rastrollo M. Ultra-Processed Food Consumption and the Incidence of Hypertension in a Mediterranean Cohort: The Seguimiento Universidad de Navarra Project. American Journal of Hypertension. 2017;30(4):358-66.
21. Rauber F, Louzada MLD, Steele EM, Millett C, Monteiro CA, Levy RB. Ultra-Processed Food Consumption and Chronic Non-Communicable Diseases-Related Dietary Nutrient Profile in the UK (2008-2014). Nutrients. 2018;10(5).
22. Monteiro EMS, Larissa Galastri B, Maria Laura da Costa L, Jean-Claude M, Dariush M, Carlos A. Ultra-processed foods and added sugars in the US diet: evidence from a nationally representative cross-sectional study. BMJ. 2016;6(3).
23. SACN. Carbohydrates and Health. London: TSO; 2015.
24. Moubarac J-C, Martins APB, Claro RM, Levy RB, Cannon G, Monteiro CA. Consumption of ultra-processed foods and likely impact on human health. Evidence from Canada. Public Health Nutrition. 2013;16(12):2240-8.
25. Monteiro CA, Levy RB, Claro RM, de Castro IRR, Cannon G. Increasing consumption of ultra-processed foods and likely impact on human health: evidence from Brazil. Public Health Nutrition. 2010;14(1):5-13.
26. Fardet A. Minimally processed foods are more satiating and less hyperglycemic than ultra-processed foods: a preliminary study with 98 ready-to-eat foods. 2016.
27. Drewnowski A, Mennella JA, Johnson SL, Bellisle F. Sweetness and Food Preference. J Nutr. 2012;142(6):1142S-8S.
28. Newman LP, Bolhuis DP, Torres SJ, Keast RS. Dietary fat restriction increases fat taste sensitivity in people with obesity. Obesity (Silver Spring). 2016;24(2):328-34.
29. Blundell JE, Bellisle F. Satiation, satiety and the control of food intake: theory and practice. Cambridge: Woodhead Publishing Ltd; 2013.
30. Roberts C, Steer T, Maplethorpe N, Cox L, Meadows S, Nicholson S, et al. National Diet and Nutrition Survey. Results from Years 7 and 8 (combined) of the Rolling Programme (2014/2015 - 2015/2016). London; 2018.
31. SACN. Folate and Disease Prevention. London: TSO; 2006.
32. Scrinis G, Monteiro CA. Ultra-processed foods and the limits of product reformulation. Public Health Nutrition. 2018;21(1):247-52.
33. Lam MCL, Adams J. Association between home food preparation skills and behaviour, and consumption of ultra-processed foods: Cross-sectional analysis of the UK National Diet and nutrition survey (2008-2009). International Journal of Behavioral Nutrition and Physical Activity. 2017;14.
34. Malam S, Prior G, Phillips R, O'Driscoll C. Food and You Survey. English Bulletin 3: Eating Outside the Home.; 2014.